NVIDIA Inception Awards Unveiled: Subtle Medical Wins Healthcare Category, Three MedTech Startups Reach Semi-Finals

VCBeat learned from the recent GTC conference that NVIDIA announced the winners of three categories in the “NVIDIA Startup Accelerator Program Challenge Awards,” who will share a $1 million prize pool.

These include: Subtle Medical in healthcare, which aims to improve medical imaging for better acquisition, reconstruction, processing, and analysis; Kinema Systems in autonomous systems, building robotic solutions for logistics and manufacturing; and AiFi at the enterprise level, developing cashier-less self-checkout systems for stores of all sizes.

The competition’s judging panel remains an all-star lineup, including Pawan Tewari of Goldman Sachs, Steve Wymer of Fidelity Investments, Jaimin Rangwalla of Coatue Management, and Jeff Herbst, Vice President of Business Development at NVIDIA.

The NVIDIA Inception Challenge attracted over 2,000 AI startups. NVIDIA categorized them into three groups, with a total of 12 ventures advancing to the semifinals. Four of these were in the healthcare category: Subtle Medical, Cambridge Bio-Augmentation Systems, Parabricks, and Phenomic AI.

After the competition, Subtle Medical won first place, and Cambridge Bio-Augmentation Systems took second. VCBeat conducted a comprehensive review of these healthcare companies.

Enhao Gong, founder of Subtle Medical, believes that medical imaging examinations such as MRI and PET take too long and are expensive. Each test takes approximately 45 minutes, a duration that has not significantly changed in recent years, with costs approaching $2,000. Part of the reason for this high price is that the lengthy testing time makes it difficult to perform enough scans in a single day to spread out fixed costs.

Subtle Medical has introduced a method that accelerates MRI scans by two to four times while maintaining image quality. The approach leverages deep learning neural networks to expedite traditional, slow, and costly examinations.

This technology can also shorten the duration of PET scans and reduce radiation doses, thereby minimizing harm to patients.

Co-founder Greg Zaharchuk stated, “Many patients cannot tolerate a 45-minute scan, particularly those with claustrophobia or pediatric patients. Our goal is to reduce the scan time by two- to four-fold, enabling technologists to image more patients per day while lowering radiation dose, all while ensuring superior diagnostic quality and greater diagnostic efficiency.”

Zaharchuk said, “We believe Subtle Medical will become a disruptive force in radiology.”

“Subtle Medical is not attempting to replace radiologists, and the system is projected to account for only 10% of the total cost of the medical imaging workflow,” emphasized Enhao Gong.

Subtle Medical, founded in August 2017, has a team of fewer than 15 employees. The company has secured $1.1 million in funding from investors including ZhenFund, Baidu Venture Capital, Data Collective, Qingyuan Capital, and Wisemont Capital.

Above: Emil Hewage, CEO of Cambridge Bioenhancement Systems

Image source: Dean Takahashi

Emil Hewage, founder and CEO of Cambridge Bio-Augmentation Systems, is developing neural interfaces for the next generation of AI-powered healthcare. The company initially focused on assisting individuals with limb loss through a project called “Neuroengineering.”

It is achieved by creating open-standard hardware and software interfaces for the nervous system, akin to equipping the body with USB ports.

“When we consider edge computing, we think about how to connect artificial intelligence to the body via the nervous system and how to change people’s lives,” said Hewage. “It is very difficult, but worth solving.”

Hewage considers the nervous system to be “the internet of our body.” He stated, “If your body loses control over the nervous system, it loses the ability to regulate organs or limbs.” What happens to neural pathways before nervous system failure occurs, such as in heart failure or during a surge in blood glucose levels?

The company is developing software that can monitor human physiological signals tens of thousands of times per second. This software aims to decipher the “biological code.”

Studies have shown that the body transmits electrical impulses at varying data rates. The spinal cord functions like a data highway, operating at a speed of 1,000 megabits per second. Organs operate at 4 megabits per second, while limbs operate at less than 1 megabit per second. A single implant can generate approximately 5 TB of data per week for the analysis of neural signals.

Hewage stated that the company collaborates with surgeons to focus on serving amputee patients, ensuring that surgical implants (such as prosthetic arms) function properly after attachment. The company has developed a neural network that identifies and interprets neural signals, translating them into interpretable neural codes. Clinicians can leverage this information to understand the interaction between prosthetic limbs and the body.

Approximately seven years ago, Hewage began conducting research in this field. Cambridge Bio-Augmentation Systems was established in the United Kingdom in May 2015 and currently employs fewer than 15 people. The company has not disclosed its funding sources.

Mehrzad Samadi, CEO of Parabricks, aims to accelerate genomics processing. Genomic information can help physicians understand a patient’s family medical history, thereby enabling them to better assess an individual’s likelihood of developing certain diseases.

As early as 2003, when the first human genome was sequenced, it took 15 years and cost $2.7 billion. Today, it requires just one day and $1,000. However, demand is rising; Samadi predicts that by 2026, companies will need to sequence the genomes of 1 billion people. Using current methods, this would cost approximately $20 billion.

“Genomics is advancing rapidly,” said Samadi. “We have mastered faster and more cost-effective methods, but the bottleneck has now shifted to another area.”

Each patient generates more than 300 gigabytes of data, and processing this data using central processing units (CPUs) takes a considerable amount of time. Samadi stated that Parabricks can accelerate the workflow and reduce computational costs by 50%, thereby significantly decreasing the number of servers required as well as the error rate.

The company employs two sales models: one is offline servers built on NVIDIA’s platform, with a one-year service term; the other is real-time services provided via the cloud.

“Parabricks delivers results in minutes, breaking the current genomics analysis market norm of waiting days,” said Samadi.

Parabricks was founded in Michigan in 2015. Currently, it has fewer than 15 employees and has raised $400,000 from the National Science Foundation, the Michigan Economic Development Corporation, and Monroe-Brown.

Phenomic AI Leverages Artificial Intelligence to Combat Cancer, Focusing on Research into Chemotherapy Resistance.

To address the issue of chemotherapy resistance, clinicians administer chemotherapy to eliminate susceptible cell populations, thereby enriching the proportion of drug-resistant cells. Subsequently, they seek agents that selectively target these resistant populations. However, a key challenge lies in the uncertainty that the surviving cells post-chemotherapy genuinely represent a pre-existing drug-resistant subpopulation.

Phenomic predicts which cells can survive chemotherapy and identifies compounds that selectively target these drug-resistant cells. It then develops these compounds and brings them to market. During the prediction process, Phenomic employs cell imaging technologies, with artificial intelligence (AI) serving as a core component of its technology stack.

“We collected data to see whether cells could survive chemotherapy,” Cooper said.

Phenomic is studying a population of chemotherapy-resistant cells. Phenomic AI was founded in Canada in June 2017. It has five employees and raised $500,000 from undisclosed investors.

Last year’s 14 award-winning and nominated companies have secured a total of $180 million in funding from investors including Sequoia Capital, Data Collective, Khosla Ventures, and Lux Capital.

Article Information Source

https://venturebeat.com/2018/03/11/nvidia-inceptions-ai-health-care-startups-cover-neural-interfaces-to-better-mri/

NVIDIA Official Website, NVIDIA Official WeChat Account, Subtle Medical Official Website